KafkaAvroDeserializer failing with Kyro Exception

问题

I have written a consumer to read Avro's generic record using a schema registry.

FlinkKafkaConsumer010 kafkaConsumer010 = new FlinkKafkaConsumer010(KAFKA_TOPICS,
                new KafkaGenericAvroDeserializationSchema(schemaRegistryUrl),
                properties);

And the Deserialization class looks like this :

public class KafkaGenericAvroDeserializationSchema implements KeyedDeserializationSchema<GenericRecord> {

   private final String registryUrl;
    private transient KafkaAvroDeserializer inner;

    public KafkaGenericAvroDeserializationSchema(String registryUrl) {
        this.registryUrl = registryUrl;
    }

    @Override
    public GenericRecord deserialize(byte[] messageKey, byte[] message, String topic, int partition, long offset) {
        checkInitialized();
        return (GenericRecord) inner.deserialize(topic, message);
    }

    @Override
    public boolean isEndOfStream(GenericRecord nextElement) {
        return false;
    }

    @Override
    public TypeInformation<GenericRecord> getProducedType() {
        return TypeExtractor.getForClass(GenericRecord.class);
    }

    private void checkInitialized() {
        if (inner == null) {
            Map<String, Object> props = new HashMap<>();
            props.put(AbstractKafkaAvroSerDeConfig.SCHEMA_REGISTRY_URL_CONFIG, registryUrl);
            props.put(KafkaAvroDeserializerConfig.SPECIFIC_AVRO_READER_CONFIG, false);
            SchemaRegistryClient client =
                    new CachedSchemaRegistryClient(
                            registryUrl, AbstractKafkaAvroSerDeConfig.MAX_SCHEMAS_PER_SUBJECT_DEFAULT);
            inner = new KafkaAvroDeserializer(client, props);
        }
    }
}

It's working locally on my machine but when I deployed it on yarn cluster I am getting below exception:

java.lang.Exception: org.apache.flink.streaming.runtime.tasks.ExceptionInChainedOperatorException: Could not forward element to next operator
    at org.apache.flink.streaming.runtime.tasks.SourceStreamTask$LegacySourceFunctionThread.checkThrowSourceExecutionException(SourceStreamTask.java:212)

Caused by: org.apache.flink.streaming.runtime.tasks.ExceptionInChainedOperatorException: Could not forward element to next operator
    at org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.pushToOperator(OperatorChain.java:654)
    at org.apache.flink.streaming.runtime.tasks.OperatorChain$CopyingChainingOutput.collect(OperatorChain.java:612)

Caused by: com.esotericsoftware.kryo.KryoException: Error constructing instance of class: org.apache.avro.Schema$LockableArrayList
Serialization trace:
types (org.apache.avro.Schema$UnionSchema)
schema (org.apache.avro.Schema$Field)
fieldMap (org.apache.avro.Schema$RecordSchema)
schema (org.apache.avro.generic.GenericData$Record)

Caused by: java.lang.IllegalAccessException: Class com.twitter.chill.Instantiators$$anonfun$normalJava$1 can not access a member of class org.apache.avro.Schema$LockableArrayList with modifiers "public"

Please help me to resolve this issue.

回答1:

As written on the mailing list:

The issue is that your are not providing any meaningful type information, so that Flink has to resort to Kryo. You need to extract the schema during query compilation (in your main) and pass it to your deserialization schema.

public TypeInformation<T> getProducedType() {
      return (TypeInformation<T>) new GenericRecordAvroTypeInfo(this.schema);
}

If you don't want to extract it statically you need to tell Flink how to handle arbitrary GenericRecords. You could implement your own serializer, which would write GenericRecords to byte[] and vice versa.

Note that I still recommend to just bundle the schema with your Flink application and not reinvent the wheel.

来源：https://stackoverflow.com/questions/59870488/kafkaavrodeserializer-failing-with-kyro-exception

标签

java

apache-kafka

apache-flink

avro

flink-streaming